git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
* (Some?) control codes not escaped when using `git add -p`
@ 2017-10-12 17:59 Mahmoud Al-Qudsi
  2017-10-12 18:47 ` Jeff King
  0 siblings, 1 reply; 2+ messages in thread
From: Mahmoud Al-Qudsi @ 2017-10-12 17:59 UTC (permalink / raw)
  To: git, Mahmoud Al-Qudsi

Hello list,

Running git 2.7.4, I’ve run into an issue where control codes that would
normally be escaped when using `git diff` are not similarly escaped when using
`git add -p`.

As a concrete example, I have a text file including the following "text":

:map ^[[H <Home>
:map ^[[5~ ^B "page up
:map ^[[6~ ^F "page down

Except each ^x above is the literal ctrl+x (i.e. ctrl+v followed by ctrl+x).
These are not lines that have been added or removed from the document, they're
just context lines.

Using `git diff`, these special characters are elided from the diff output
(though the latter two lines cause special coloring in the diff output so
perhaps they're not being entirely escaped?), but when using `git add -p` upon
reaching the chunk in question my terminal interprets a literal "page up" input
and corrupts the output.

Here's a screenshot of what I see when I use `git diff`:
https://i.imgur.com/FOXWEIi.png

And here's what I see when use `git add -p`:
https://i.imgur.com/i5hqhFX.png

As you can see, in the second example the cursor is a few lines from the top of
the screen, as the text output started midway down and then jumped to the start
and continued from there on encountering the literal 'Page Up' sequence in the
diff'd text.

I'm not sure _what_ the correct approach would be here (does git faithfully
print whatever it finds in the file or does it attempt to escape it instead?)
but it seems to me that the lack of consistency between the two commands is a
bug as whichever approach is taken, a context line in `git diff` should surely
be output to the terminal in the same way as a context line in `git add -p`?

The obvious solution is to embrace isatty(2) religiously, but I'm not
sure how the
everyone else feels about that (or if it's already used elsewhere).

Anyway, I'm sure I'm not the only one to run into this. Seeking options and
interested in the various viewpoints on how this should be correctly handled
(or explanations for why it's already correct as-is).

Cheers,

Mahmoud Al-Qudsi
NeoSmart Technologies

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2017-10-12 18:47 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-10-12 17:59 (Some?) control codes not escaped when using `git add -p` Mahmoud Al-Qudsi
2017-10-12 18:47 ` Jeff King

Code repositories for project(s) associated with this public inbox

	https://80x24.org/mirrors/git.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).